CERN Accelerating science

A high-performance portable abstract interface for explicit SIMD vectorization

Date published: 
Saturday, 4 February, 2017
Document type: 
Conference paper
P. Karpinski, J. McDonald
Œis work establishes a scalable, easy to use and ecient approach for exploiting SIMD capabilities of modern CPUs, without the need for extensive knowledge of architecture specifi€c instruction sets. We provide a description of a new API, known as UME::SIMD, which provides a flexible, portable, type-oriented abstraction for SIMD instruction set architectures. Requirements for such libraries are analysed based on existing, as well as proposed future solutions. A so‰ware architecture that achieves these requirements is explained, and its performance evaluated. Finally we discuss how the API €ts into the existing, and future so‰ware ecosystem.
Event published at: 
PMAM'17 Proceedings of the 8th International Workshop on Programming Models and Applications for Multicores and Manycores
Technical document file: